J. Dairy Sci. 87:386–398 American Dairy Science Association, 2004. Evaluation of Milk Urea Nitrogen as a Diagnostic of Protein Feeding J. Nousiainen, 1 K. J. Shingfield, 2,3 and P. Huhtanen 2 1 Valio Ltd, Farm Services, P.O. Box 10, FIN-00039 Valio, Finland 2 MTT Agrifood Research Finland, Animal Production Research, FIN-31600 Jokioinen, Finland 3 School of Food Biosciences, The University of Reading, PO Box 226, Reading, RG6 6AP, UK ABSTRACT An evaluation of milk urea nitrogen (MUN) as a diag- nostic of protein feeding in dairy cows was performed using mean treatment data (n = 306) from 50 production trials conducted in Finland (n = 48) and Sweden (n = 2). Data were used to assess the effects of diet composi- tion and certain animal characteristics on MUN and to derive relationships between MUN and the efficiency of N utilization for milk production and urinary N excre- tion. Relationships were developed using regression analysis based on either models of fixed factors or using mixed models that account for between-experiment variations. Dietary crude protein (CP) content was the best single predictor of MUN and accounted for propor- tionately 0.778 of total variance [MUN (mg/dL) = −14.2 + 0.17 × dietary CP content (g/kg dry matter)]. The proportion of variation explained by this relationship increased to 0.952 when a mixed model including the random effects of study was used, but both the intercept and slope remained unchanged. Use of rumen degrad- able CP concentration in excess of predicted require- ments, or the ratio of dietary CP to metabolizable en- ergy as single predictors, did not explain more of the variation in MUN (R 2 = 0.767 or 0.778, respectively) than dietary CP content. Inclusion of other dietary fac- tors with dietary CP content in bivariate models re- sulted in only marginally better predictions of MUN (R 2 = 0.785 to 0.804). Closer relationships existed be- tween MUN and dietary factors when nutrients (CP to metabolizable energy) were expressed as concentra- tions in the diet, rather than absolute intakes. Further- more, both MUN and MUN secretion (g/d) provided more accurate predictions of urinary N excretion (R 2 = 0.787 and 0.835, respectively) than measurements of the efficiency of N utilization for milk production (R 2 = 0.769). It is concluded that dietary CP content is the most important nutritional factor influencing MUN, Received May 14, 2003. Accepted July 15, 2003. Corresponding author: P. Huhtanen; e-mail: pekka.huhtanen@ mtt.fi. 386 and that measurements of MUN can be utilized as a diagnostic of protein feeding in the dairy cow and used to predict urinary N excretion. (Key words: milk urea nitrogen, dairy cow, protein nutrition, diagnostic) Abbreviation key: AAT = amino acids absorbed from the small intestine, BUN = blood urea nitrogen, ME = metabolizable energy, ECM = energy-corrected milk, MUNS = milk urea N secretion, PBV = protein balance in the rumen, RMSE = residual mean square error. INTRODUCTION The basic function of milk producing ruminants is to convert low-quality noncompetitive feed sources into high quality protein for human consumption. Often the amount and quality of protein absorbed from the small intestine can limit milk production (Huhtanen, 1998). However, feeding excess protein in relation to require- ments increase environmental N emissions (Castillo et al., 2000; Frank and Swensson, 2002; Huhtanen et al., unpublished) and can impair reproductive performance (refer to Shingfield et al., 1999). Consequently, there is an urgent need for on-farm diagnostic to monitor the adequacy of protein feeding offering the opportunity to optimize the efficiency of N utilization with respect to both milk protein production and N emissions into the environment. Blood urea nitrogen (BUN) is the major end product of N metabolism in ruminants, and high concentrations of it are indicative of an inefficient utilization of dietary N. However, BUN cannot be measured routinely due to difficulties in obtaining regular and reliable samples. It is well established that urea equilibrates rapidly with body fluids, including milk, and this can account for the close relationship between milk urea nitrogen (MUN) and BUN (Broderick and Clayton, 1997; Hof et al., 1997). Since milk is easily collected and can be determined accurately for urea by enzymatic or physi- cal methods, it has often been suggested that MUN in bulk tank milk could be used as a diagnostic of on-farm efficiency of N utilization (Jonker et al., 1998; Kauffman and St-Pierre, 2001; Kohn et al., 2002).
Transcript
J. Dairy Sci. 87:386–398 American Dairy Science Association, 2004.
Evaluation of Milk Urea Nitrogen as a Diagnostic of Protein Feeding
J. Nousiainen,1 K. J. Shingfield,2,3 and P. Huhtanen21Valio Ltd, Farm Services, P.O. Box 10, FIN-00039 Valio, Finland2MTT Agrifood Research Finland, Animal Production Research,FIN-31600 Jokioinen, Finland3School of Food Biosciences, The University of Reading,PO Box 226, Reading, RG6 6AP, UK
ABSTRACT
An evaluation of milk urea nitrogen (MUN) as a diag-nostic of protein feeding in dairy cows was performedusing mean treatment data (n = 306) from 50 productiontrials conducted in Finland (n = 48) and Sweden (n =2). Data were used to assess the effects of diet composi-tion and certain animal characteristics on MUN and toderive relationships between MUN and the efficiencyof N utilization for milk production and urinary N excre-tion. Relationships were developed using regressionanalysis based on either models of fixed factors or usingmixed models that account for between-experimentvariations. Dietary crude protein (CP) content was thebest single predictor of MUN and accounted for propor-tionately 0.778 of total variance [MUN (mg/dL) = −14.2+ 0.17 × dietary CP content (g/kg dry matter)]. Theproportion of variation explained by this relationshipincreased to 0.952 when a mixed model including therandom effects of study was used, but both the interceptand slope remained unchanged. Use of rumen degrad-able CP concentration in excess of predicted require-ments, or the ratio of dietary CP to metabolizable en-ergy as single predictors, did not explain more of thevariation in MUN (R2 = 0.767 or 0.778, respectively)than dietary CP content. Inclusion of other dietary fac-tors with dietary CP content in bivariate models re-sulted in only marginally better predictions of MUN(R2 = 0.785 to 0.804). Closer relationships existed be-tween MUN and dietary factors when nutrients (CP tometabolizable energy) were expressed as concentra-tions in the diet, rather than absolute intakes. Further-more, both MUN and MUN secretion (g/d) providedmore accurate predictions of urinary N excretion (R2 =0.787 and 0.835, respectively) than measurements ofthe efficiency of N utilization for milk production (R2 =0.769). It is concluded that dietary CP content is themost important nutritional factor influencing MUN,
Received May 14, 2003.Accepted July 15, 2003.Corresponding author: P. Huhtanen; e-mail: pekka.huhtanen@
mtt.fi.
386
and that measurements of MUN can be utilized as adiagnostic of protein feeding in the dairy cow and usedto predict urinary N excretion.(Key words: milk urea nitrogen, dairy cow, proteinnutrition, diagnostic)
Abbreviation key: AAT = amino acids absorbed fromthe small intestine, BUN = blood urea nitrogen, ME =metabolizable energy, ECM = energy-corrected milk,MUNS = milk urea N secretion, PBV = protein balancein the rumen, RMSE = residual mean square error.
INTRODUCTION
The basic function of milk producing ruminants is toconvert low-quality noncompetitive feed sources intohigh quality protein for human consumption. Often theamount and quality of protein absorbed from the smallintestine can limit milk production (Huhtanen, 1998).However, feeding excess protein in relation to require-ments increase environmental N emissions (Castillo etal., 2000; Frank and Swensson, 2002; Huhtanen et al.,unpublished) and can impair reproductive performance(refer to Shingfield et al., 1999). Consequently, there isan urgent need for on-farm diagnostic to monitor theadequacy of protein feeding offering the opportunity tooptimize the efficiency of N utilization with respect toboth milk protein production and N emissions intothe environment.
Blood urea nitrogen (BUN) is the major end productof N metabolism in ruminants, and high concentrationsof it are indicative of an inefficient utilization of dietaryN. However, BUN cannot be measured routinely dueto difficulties in obtaining regular and reliable samples.It is well established that urea equilibrates rapidly withbody fluids, including milk, and this can account forthe close relationship between milk urea nitrogen(MUN) and BUN (Broderick and Clayton, 1997; Hof etal., 1997). Since milk is easily collected and can bedetermined accurately for urea by enzymatic or physi-cal methods, it has often been suggested that MUN inbulk tank milk could be used as a diagnostic of on-farmefficiency of N utilization (Jonker et al., 1998; Kauffmanand St-Pierre, 2001; Kohn et al., 2002).
MILK UREA AS A PROTEIN DIAGNOSTIC 387
Variance in MUN has been shown to be related tothe ratio of dietary CP to energy (Oltner and Wiktors-son, 1983; Kirchgessner et al., 1986), extent of CP deg-radation in the rumen and the amount of ammonia inexcess of microbial N requirements (Ropstad et al.,1989; Roseler et al., 1993; Hof et al., 1997), and proteinor energy intake in relation to feeding standards (Gus-tafsson and Carlsson, 1993; Carlsson and Pehrson,1994). However, there is evidence that MUN is moreclosely associated with changes in dietary CP contentthan the ratio of dietary CP to energy intake, efficiencyof N utilization, or rumen ammonia concentration(Broderick and Clayton, 1997). Even though these find-ings have been based on sound physiological principles,the relationships derived may not be universally appli-cable due to variations in nutrient intake and between-animal differences. Furthermore, previous evaluationsof MUN fail to account for random study effects, whichcan lead to biased estimates of regression coefficients(St-Pierre, 2001) and have been based on data from alimited number of studies (e.g., Jonker et al., 1998;Kauffman and St-Pierre, 2001; Kohn et al., 2002).
Milk urea nitrogen has routinely been determined ina number of milk production trials conducted in Nordiccountries that also measure animal and dietary charac-teristics. This evaluation was conducted to (1) examinethe effects of diet composition and nutrient intake onMUN using mean treatment data across a wide rangeof nutritional circumstances, in conjunction with statis-tical models that accounted for random study effects,and (2) assess the potential of MUN as a predictor ofurinary N excretion and the efficiency of dietary N utili-zation.
MATERIALS AND METHODS
Datasets
Mean treatment data were derived from 50 milk pro-duction trials that assessed 306 different diets. Trialswere conducted in Finland (n = 48) and Sweden (n =2; refer to Appendix 1), most of which (n = 42) wereconducted as changeover designs with 3- or 4-wk experi-mental periods.
Experimental Diets
Grass or grass-legume silages were fed ad libitum inall trials supplemented with concentrates offered at aflat rate (mean 8.1 kg DM/d, SE 1.26), irrespective ofmilk yield. For statistical analysis, diet composition wascharacterized as (g/kg DM) CP, AA absorbed from thesmall intestine (AAT), protein balance in the rumen(PBV), NDF, starch, NSC, lactic acid, VFA, and ammo-nia N. Apparent diet digestibility was determined (n =
Journal of Dairy Science Vol. 87, No. 2, 2004
270) using acid insoluble ash as an internal marker(Van Keulen and Young, 1977). Nonstructural carbohy-drate content was estimated as OM − (CP + ether ex-tract + NDF). Concentrations of AAT and PBV werecalculated according to the metabolizable protein sys-tem adopted in Finland (Tuori et al., 2002), where AATrepresents the supply of AA available for absorption,and PBV indicates the balance of rumen degradablecrude protein relative to microbial N requirements, pro-viding an estimate of rumen N losses. Metabolizableenergy (ME) content was calculated using publishedvalues for concentrate ingredients (Tuori et al., 2002),and in vivo or in vitro OM digestibility of forages. Invivo OM digestibility of silages was measured in sheepfed at maintenance (n = 13), or predicted in vitro usingrumen fluid (n = 16, Tilley and Terry, 1963) or a pepsin-cellulase (n = 25) based method (see Nousiainen etal., 2003).
In addition to diet composition, including silage, con-centrate, and total DMI, animal measures such as dailymilk and energy-corrected milk (ECM) yield, milk pro-tein, fat and lactose concentration, live weight, andmean DIM during experiment were used. Milk ureanitrogen was determined as ammonia in compositesamples according to McCullough (1967) (26 trials and185 treatments), or Rajamaki and Rauramaa (1984) (8trials and 39 treatments), or measured in untreatedsamples by an automated infrared analysis (MilcoscanIR 605 or IR 4000, FOSS Electric A/S, Denmark; 16trials and 82 treatments).
The apparent efficiency of N utilization for milk pro-tein synthesis was defined as milk N/N intake. UrinaryN excretion was estimated as N intake − (milk N + fecalN output), assuming no net changes in N retentionduring each experiment. To estimate the source of MUNsecretion in milk, the contribution of absorbed N notincorporated in milk protein was separated into rumi-nal N losses and those associated with the metabolismof absorbed AA (maintenance, milk production, and re-tention). In cases where PBV intake was positive, Nsurplus to microbial requirements was assumed to becompletely absorbed from the rumen and converted tourea. For diets in which PBV was negative, N lossesassociated with AA metabolism were assumed to berecycled via saliva into the rumen and to satisfy micro-bial requirements. Nitrogen losses related to the metab-olism of absorbed AA were calculated as 0.16 × AATintake (g/d) − milk N output (g/d).
Statistical Analysis
Relationships between MUN with animal and dietaryfactors were derived using the entire data or subsets,in which data were separated into studies comparing
NOUSIAINEN ET AL.388
effects of the level of concentrate supplementation (35comparisons, 80 diets), concentrate CP content (76 com-parisons, 188 diets), silage fermentation quality (21comparisons, 55 diets), or replacing grass silage withlegume silage (10 comparisons, 32 diets). In studiesevaluating the effects of concentrate feeding, only thelevel of supplementation was altered (range 2.8 to 13.5kg DM/d). For protein feeding studies, concentrateswere fed at a fixed rate (mean 7.9 kg DM/d, SE 1.55),and protein supplements replaced basal concentrateingredients. In studies assessing the impact of silagefermentation, experimental silages were prepared fromthe same sward and ensiled using none, an enzymeand/or inoculant, or formic acid based additive.
Relationships between MUN and animal or dietarymeasurements were estimated by linear regressionanalysis. Furthermore, MUN was used to predict theefficiency of N utilization for milk production and uri-nary N excretion. Relationships obtained were assessedbased on the proportion of variance accounted for bythe model (R2) and residual mean square error (RMSE).
Since a component of the variance in MUN can arisefrom differences in the stage of lactation, genetic merit,animal live weight, feeding strategies, and analyticaltechniques used to determine urea in milk, it is im-portant to exclude these sources when the impact ofnutrition is being evaluated. Therefore relationshipsbetween MUN with N utilization, animal, or dietaryfactors within the experiment were also investigatedusing the MIXED procedure of SAS (Littell et al., 1996)according to the following model Yij = A0 + Expi + B1X1ij+ B2X2ij + … + eij, where A0 is the overall intercept(fixed effect), Expi is the random effect of experiment,B1 and B2, are the overall regression coefficients acrossexperiments, X1ij and X2ij, are the value j of the continu-ous variables X1 and X2, in experiment i, and eij repre-sents unexplained error. In addition, some analysis wasconducted using random slopes to assess the extent ofRMSE attributable to the experiment. The mixed modelmethodology used has been described in detail (St-Pierre, 2001), whereas RMSE and the coefficient of de-termination (R2) were derived from single factor regres-sion between values predicted by the mixed model andmeasured values. Furthermore, R2 values were alsocalculated to estimate the proportion of variation ex-plained within the experiment, where: R2
within = [(R2 −R2
explained by experiment)/(1 − R2explained by experiment)].
RESULTS
Diet Composition
A description of the entire data used for evaluationis presented in Table 1. Most variables exhibited large
Journal of Dairy Science Vol. 87, No. 2, 2004
variance that was normally distributed. On average,cows were 109 DIM (range 42 to 226) and of mean579 kg of live weight (range 468 to 657). Diets werepredominantly based on grass silage supplementedwith a wide range of energy (cereals and cereal byprod-ucts) and protein (mainly rapeseed meal) supplements.Diets contained mean CP and ME concentrations of 160g/kg DM (range 111 to 249, coefficients of variation11.9%), and 11.3 MJ/kg DM (range 9.6 to 12.2, CV4.2%), respectively.
Feed Intake, Milk Production, and Composition
Total DMI was on average 19.8 kg/d but varied be-tween 12.9 and 23.6 kg/d, and mean CP intake was3167 g/d (range 1697 to 4903, proportionately 0.559from forage and 0.441 from concentrate CP). Both ap-parent total tract OM (mean 0.716) and N digestibility(mean 0.685) were typical for diets based on high qual-ity forages supplemented with modest levels of concen-trates. Mean daily milk and ECM yield were 27.8 (range13.0 to 36.3) and 29.0 kg/d (range 15.6 to 38.1), respec-tively, whereas MUN (mean 13.3 mg/dL) varied (CV26.0%) between 3.8 and 27.0 mg/dL.
Relationships with MUN DerivedUsing the Entire Dataset
Both fixed and mixed regression models indicated aclose association between dietary CP and MUN (Figure1, Table 2). Slopes based on single factor regression(0.172), or using mixed models that assumed randomintercepts (0.169), or both random intercepts and slopes(0.165) were similar. However, R2 values of regressionswere much higher using mixed models. Inclusion ofdietary energy content in addition to CP within bivari-ate models resulted in only marginal improvements inR2 values compared with dietary CP content alone.Overall, bivariate models indicated that at constantdietary CP content, increases in dietary NDF contentwere associated with marginally higher MUN (slope0.007 mg/dL per 1 g/kg DM increase in NDF), whereasincreases in dietary NSC content were associated withslight decreases in MUN (slope −0.007 mg/dL per 1 g/kg DM increase in NSC).
The relationship between MUN and dietary PBV con-tent was similar to that for dietary CP concentrations(R2
within 0.860 vs. 0.825), but the slope was higher forPBV (0.21 and 0.17 mg/dL per g/kg DM, respectively),and the intercept suggested that when PBV was zero(i.e., no net absorption of, ammonia from the rumen)MUN was close to 12 mg/dL. Expressing dietary CP asa ratio to energy content (CP/ME) did not improve upon
MILK UREA AS A PROTEIN DIAGNOSTIC 389
Table 1. Description of diet composition, nutrient intake, and milk production data used for evaluation.
1AAT = Amino acids absorbed from the small intestine.2PBV = Protein balance in the rumen.3ME = Metabolizable energy.4MUN = Milk urea nitrogen.
the relationship obtained with dietary CP concen-tration.
The relationship with MUN and CP intake was muchweaker than for dietary CP content based on singleregression analysis (R2 = 0.451 vs. 0.778) or using mixedmodels (R2
within 0.663 vs. 0.860). Bivariate models thatincluded DM or ME intake in addition to CP intakeexplained much more of the variation in MUN thanCP intake alone (R2
within = 0.871 and 0.861 vs. 0.663,respectively). Use of a more theoretically sound descrip-tion of dietary protein in terms of PBV and AAT, orboth these parameters and calculated energy balance,did not improve upon these relationships. Milk ureanitrogen was increased in response to both increasedAAT and PBV intake, but the slope for PBV was mark-edly higher than for AAT when both effects were usedin the same mixed model (0.002 vs. 0.011 mg/dL pergram AAT or PBV). However, assuming zero changes
Journal of Dairy Science Vol. 87, No. 2, 2004
in N-retention and expressing surplus dietary N fromruminal (PBV-N) or tissue metabolism (AAT-N − milkN) on a DMI basis (g N/kg DM) indicated that bothsources had similar effects on MUN (1.19 vs. 1.21 mg/dL per gram N/kg DMI, data not presented).
Use of the mixed effect model (Table 3) indicated thatthe intake of rumen degradable N in excess of microbialrequirements (PBV-N) increased MUN secretion morethan absorbed AA-N intake (0.018 vs. 0.012 g/g N in-take; P < 0.001), a finding that also held true using thefixed effect model. However, the impact of N losses onMUN secretion due to ammonia N absorption from therumen or AAT-N intake in excess of requirements(AAT-N intake − milk protein N output) were of simi-lar magnitude.
Silage fermentation characteristics had minor, albeit,significant effects on MUN (data not shown). Mixedmodel analysis indicated that MUN increased by 0.17
NOUSIAINEN ET AL.390
Figure 1. Association between dietary CP content and MUN forthe entire evaluation dataset (n = 306).
(P < 0.001) or 0.04 mg/dl (P < 0.01) per g/kg DM increasein dietary VFA or lactic acid content, respectively. Inaddition, MUN was also increased 0.03 mg/L (P = 0.03)in response to 1 g/kg increases in silage ammonia N con-centrations.
The outcome of the mixed effect (random intercepts)model including variables that significantly (P < 0.10)explained variation in MUN is shown in Table 4. Over-all, the model accounted for proportionally 0.957 of totalvariance, and 0.901 of within experiment variation inMUN. Dietary CP content was the most important nu-tritional factor in the model. Both dietary energy con-tent and the ratio of AAT to CP were inversely relatedwith MUN, whereas starch and NDF concentrationshad slightly positive effects. Concentrations of urea alsoincreased (P = 0.04) when milk protein content de-creased and for silages prepared from sward regrowthsrelative to the corresponding primary growths (P =0.008). Use of a formic acid based additive during ensil-ing was associated with a decrease in MUN comparedwith silages conserved using none or enzyme and/orinoculant additives. Inclusion of energy balance or milkyield as fixed factors did not account for more of thevariation in MUN, and the respective slopes were notsignificant (P > 0.10).
Relationships with MUN DerivedUsing Data Subsets
Different data subsets used to assess the reliabilityof MUN predictions indicated that MUN responses toincreases in dietary CP concentration were much the
Journal of Dairy Science Vol. 87, No. 2, 2004
same, irrespective of dietary manipulation (Table 5).However, the regression coefficient derived betweenMUN and dietary CP concentration was marginallylower than the overall mean coefficient when proteincontent was increased through partial or complete re-placement of grass silage with legume silage (0.14 vs.0.17 mg/dL per grams of CP).
The coefficient obtained using the ratio of CP to MEwas higher for studies examining the impact of proteinsupplementation compared with other data subsets.However, relationships obtained using bivariate mod-els indicated similar coefficients for CP and ME intakes,irrespective of data source.
Nitrogen Utilization and Urinary N Excretion
A negative association existed between MUN and theefficiency of N utilization for milk protein synthesis(Table 6), but dietary CP concentration explained moreof the variation in N efficiency than MUN (R2
within 0.769vs. 0.805). For both fixed and mixed models, inclusionof milk yield improved the accuracy of MUN based pre-dictions of urinary N excretion and efficiency of N utili-zation. Effects of milk yield and MUN were combinedby calculating daily MUN secretion (MUNS, g/d), whichexplained proportionately 0.835 of within experimentvariation in urinary N excretion. Furthermore, this pre-diction was associated with an RMSE of 14.7 g N perday. For data subsets based on studies assessing con-centrate feeding (1) or protein supplementation (2), thefollowing regression equations were derived usingmixed effects models:
within 0.805, RMSE = 9.6).(2) Urinary N excretion (g/d) = 32.0 (±5.25) +
1009(±23.6) MUNS (R2within 0.945, RMSE = 7.7).
In common with N efficiency, dietary CP concentra-tion provided a more accurate prediction of urinary Nexcretion than MUN. Use of MUNS in combination withdietary CP content resulted in only marginal improve-ments in the prediction of urinary N excretion for bothfixed and random effect models (Table 6).
DISCUSSION
Physiological Basis of MUN
Milk urea nitrogen has often been used to provide anindication of the efficiency of N utilization and to predictN emissions into the environment. Since urea equili-brates rapidly between body fluids (DePeters and Fer-guson, 1992) and enters the mammary gland by diffu-sion, concentrations of urea in plasma and milk aregenerally closely associated (Oltner and Wiktorsson,
MILK UREA AS A PROTEIN DIAGNOSTIC 391
Table 2. Predictions of milk urea N (mg/dl) according to linear fixed or mixed effects regression models based on diet composition, nutrientintake, or milk yield as independent regression variables (model: Y = A + BX1 + CX2, where A is the intercept, B and C represent regressionvariables, and Exp is the random effects of experiment used in mixed effects regression models, n = 306).
R2
withinX1, X2 A SE P-value B SE P-value C SE P-value RMSE CV R2 Exp
1CP = Crude protein (g/kg DM).2ME = Metabolizable energy (MJ/kg DM).3NDF = Neutral detergent fiber (g/kg DM).4NSC = Nonstructural carbohydrates (g/kg DM).5PBV = Protein balance value (g/kg DM).6AAT = Absorbed amino acids (g/d).7Milk = Milk yield (kg/d).
1983; Roseler et al., 1993; Broderick and Clayton, 1997;Kauffman and St-Pierre, 2001). Milk urea is also de-rived from arginine catabolism in the mammary gland(Annison, 1983), but this does not appear to be quantita-tively important.
Table 3. Predictions of milk urea nitrogen secretion (g/d) based on PBV-N and AAT-N intake according to fixed or mixed effects linearregression models; Y = A + BX1 + CX2, where A is the intercept, B and C represent regression variables, and Exp is the random effect ofexperiment used in mixed effects regression models, n = 306).
R2
withinX1, X2 A SE P-value B SE P-value C SE P-value RMSE R2 Exp
1PBV-NI = Excess rumen degradable N intake (g/d).2AAT-NI = Intake of absorbed amino acid N (g/d).3Excess AAT-NI = Milk protein N − intake of absorbed amino acid N (g/d).
Journal of Dairy Science Vol. 87, No. 2, 2004
The intake of PBV had a greater effect on MUN thanAAT (Table 2), which may reflect the close associationbetween AAT and energy intake. Including ME intakeas an additional independent variable reduced the dif-ference between coefficients for AAT and PBV (MUN
NOUSIAINEN ET AL.392
Table 4. Identification of dietary and production parameters that made a significant contribution in regres-sions with milk urea nitrogen concentration (mg/dl) according to mixed effects linear regression models(experiment used as a random factor, n = 259, root mean square error = 0.68, R2 = 0.957).
Estimate StandardEffect (mg/dl) error t Value Pr > |t|
1Amino acids absorbed (g) from the small intestine per kg CP.2Metabolizable energy (MJ/kg DM).3Neutral detergent fiber (g/kg DM).4Silage harvest refers to a comparison of silages harvested from primary or corresponding regrowths of
grass or grass legume swards.5Silage additive refers to a comparison of silages prepared without additive or a biological inoculant or
formic acid based ensiling additive.
(mg/dl) = 10.4(±0.90) + 0.007(±0.00015) × AAT intake(g/d) + 0.010(±0.00037) × PBV intake (g/d) −0.047(±0.0136) × ME intake (MJ/d)). This relationshipsuggests that each gram of absorbed N derived fromAA increased MUN to a lesser extent than that fromrumen ammonia, a finding consistent with a 50% higher
Table 5. Predictions of milk urea nitrogen (mg/dl) based on mixed linear effects regression models (Y = A + BX1 + CX2) in studies wherenutrient intake was manipulated by the level of concentrate feeding, protein supplementation level, forage type (legume vs. grass silages),or silage fermentation quality. (Experiment was used as random factor.)
Dataset X1, X2 n A SE P-value B SE P-value C SE P-value RMSE CV R2
1CP = Crude protein (g/kg dry matter DM).2PBV = Protein balance value (g/kg DM).3ME = Metabolizable energy (MJ/ kg DM).4CPI = Crude protein intake (g/d).5MEI = Metabolizable energy intake (MJ/d).
Journal of Dairy Science Vol. 87, No. 2, 2004
coefficient for rumen degradable than undegraded Nintake (DePeters and Ferguson, 1992).
Prediction of MUN secretion based on PBV-N andAAT-N intake also suggests that absorption of N fromruminal ammonia has a markedly greater impact thantissue AA catabolism (Table 3). However, absorbed AA
MILK UREA AS A PROTEIN DIAGNOSTIC 393
Table 6. Predictions of the efficiency of nitrogen (N) utilization for milk protein synthesis (g/kg N intake) and urinary N excretion (g/d)based on MUN1 or daily MUN secretion, dietary CP content and milk yield according to linear fixed or mixed effects regression models;Y = A + BX1 + CX2, where A is the intercept, B and C represent regression variables, and Exp is the random effect of experiment used inmixed effects regression models, (n = 306).
not utilized for milk protein synthesis had similar ef-fects on milk urea N yield as ruminal N losses. Thisfinding is not consistent with the view that the contribu-tion of urea in biological fluids from AA absorption andsubsequent metabolism is relatively minor (Hof et al.,1997; Schepers and Meijer, 1998). Based on theoreticalconsiderations, these sources would be expected to havesimilar effects on MUN, because both ammonia N ab-sorbed from the rumen and absorbed AA N not utilizedfor milk protein synthesis, or retained in body tissues,are metabolized to urea in the liver. Furthermore, MUNhas increased in response to postruminal casein infu-sions in cows fed grass silage based diets (Huhtanenet al., 1997; Vanhatalo et al., 2003), strongly suggestingthat AA catabolism is a significant source of BUN andMUN. Consistent with regression coefficients obtainedin this evaluation, both ruminal and duodenal infusionsof casein have increased MUN, but the response to ru-minal infusions was higher (Khalili and Huhtanen,2002), due to higher milk protein secretion and smallerlosses of absorbed N.
The Effect of Diet on MUN
For both fixed and mixed models, dietary CP concen-tration was the best single predictor of MUN, which is
Journal of Dairy Science Vol. 87, No. 2, 2004
in direct contrast to a number of studies identifying thedietary ratio of CP to energy as the most importantnutritional factor affecting MUN (e.g., Oltner and Wik-torsson, 1983; Kirchgessner et al., 1986). Oltner andWiktorsson (1983) postulated that MUN was moreclosely associated with the ratio of CP to energy in thediet. However, further interpretation of mean treat-ment data of this study indicated that the relationshipbetween MUN with dietary CP content (R2 = 0.92) wasas strong as that for the CP/ME ratio (R2 = 0.92). Similarfindings can also be explained by variations in the CP/ME ratio being confounded with dietary CP concentra-tions, because increases in energy intake were associ-ated with concomitant decreases in protein supplemen-tation. Broderick and Clayton (1997) also reported mar-ginally closer relationships between MUN with dietaryCP content than the ratio of CP to net energy for lacta-tion. However, within the current dataset, differencesin dietary CP (R2 = 0.89) rather than ME content (R2 =0.03), were the major source of variation in the ratio ofdietary CP to ME. Furthermore, true variance in di-etary ME concentrations may be lower than estimatessuggest, due to negative associative effects of digestionresulting in the increase in energy content with moreintensive concentrate feeding being lower than ex-pected (Huhtanen, 1998).
NOUSIAINEN ET AL.394
It has been suggested that a restriction in energysupply increases MUN (Kirchgessner et al., 1986), butthere was no evidence of a close positive associationwith calculated energy balance based on the currentevaluation. This apparent discrepancy may reflect en-ergy balance being confounded with dietary CP contentin many studies.
Use of both dietary AAT and PBV as independentvariables in bivariate models would be expected to ac-count for more variation in MUN than dietary CP, butthis evaluation does not support this (Table 2). The lackof improvement in the prediction of MUN suggests thatthe errors in estimating rumen degradable and unde-gradable protein outweigh the potential benefits in ac-counting for different physiological sources of urea se-creted in milk. Even though AAT and PBV contentfailed to explain substantially more of the variance inMUN than CP content for the entire dataset, individualstudies have established that MUN is also related tothe quality of dietary protein. In a recent comparisonover a wide range of inclusion rates, iso-nitrogenoussupplements of heat-treated rapeseed expeller elicitedhigher milk protein and lower MUN responses com-pared with solvent-extracted soybean meal in cows fedgrass silage based diets (Shingfield et al., accepted).Similarly, provision of comparable amounts of dietaryN as rapeseed expeller enhanced milk protein outputand reduced secretion of urea in milk, compared withdietary urea supplements or the use of higher N fertil-izer application rates to increase grass silage CP con-tent (Shingfield et al., 2001).
Irrespective of the model or dataset used, the inter-cept of the regression between dietary PBV concentra-tion (or PBV intake) and MUN was approximately 11.7mg/dL, suggesting at this concentration the availabilityof degradable N and energy supply in the rumen areessentially balanced. A very similar value of 10.7 mg/dL has been derived based on the metabolizable proteinsystem adopted in the Netherlands (Hof et al., 1997).However, measurements of microbial protein enteringthe omasal canal have suggested that microbial N re-quirements can be satisfied (Ahvenjarvi et al., 1999,2002) even for diets containing negative amounts ofPBV based on the Finnish metabolizable protein system(Tuori et al., 2002).
Practical Applications of MUN
Milk urea nitrogen can be relatively easily analyzedin bulk tank or individual milk samples of herds partici-pating in herd improvement schemes (see Godden etal., 2000). This evaluation supports the suggestion thatmeasurements of MUN could be used to assess the ade-quacy of protein feeding in dairy cows and the efficiency
Journal of Dairy Science Vol. 87, No. 2, 2004
of N utilization for milk production (Broderick andClayton, 1997; Jonker et al., 1998 and 2002; Kohn etal., 2002).
Of all the components assessed, dietary CP contentwas the most influential with respect to MUN and couldtherefore be used to predict CP concentrations of dietsfed on-farm. Based on measurements from all experi-ments and diets (n = 306) dietary CP content was accu-rately predicted from MUN as: CP (g/kg DM) = 99.8(1.9)+ 4.5(±0.01) × MUN (mg/dL) (R2 = 0.779; RMSE 9.0 g/kgDM). Both the intercept (93.4) and regression coefficient(5.0) were similar when the effects of between-experi-ment variations were omitted or taken into account,whereas the RMSE of 9.0 (g/kg DM) is sufficiently smallto suggest that this equation could be used to assessprotein adequacy under most practical feeding situa-tions. In spite of a higher intercept and regression coef-ficient, this prediction is consistent with that from anearlier evaluation of the potential of MUN to predicton-farm dietary CP content (Broderick and Clayton,1997). Even though these parameters are different, dueto variations in mean dietary CP content and a tendencyfor curvilinear association with MUN at high CP con-centrations in the diet (P = 0.01 for quadratic slope indata subset of legume forages), this does not necessarilynegate the use of MUN measurements, provided thatthe prediction equation is applied to similar nutritionalcircumstances under which it was developed. For exam-ple, when comparing measured MUN concentrations(Frank and Swensson, 2002) and those estimated basedon the relationship between dietary CP and MUN (Ta-ble 2), a close correlation existed between observed andpredicted values (R2 = 0.95, RMSE = 0.74).
In addition to assessing dietary CP content, measure-ments of MUN may also yield useful information con-cerning the utilization of dietary rumen degradable pro-tein. This evaluation suggests that for diets based ongrass silage, a MUN value of 11.7 mg/dL is consistentwith the N requirements of rumen microbes being satis-fied. Because recycling of N into the rumen was notaccounted for, a lower concentration may well be ade-quate. A quadratic relationship between MUN and milkprotein yield (Figure 2) indicates that increases in milkprotein yield could be expected in response to high qual-ity protein supplements beyond MUN concentrationsof 11.7 mg/dL. Even though production responses canbe attained above MUN values of 16 mg/dL, this occursat the expense of a reduction in the efficiency of Nutilization (Figure 2). The implication is that establish-ing recommended values for MUN is heavily dependenton the criteria being considered for optimization (i.e.,recommendations for milk protein production may notnecessarily coincide with those regarded optimal with
MILK UREA AS A PROTEIN DIAGNOSTIC 395
Figure 2. Association between the efficiency of dietary N for milkprotein production [◆; y = 0.0049(±0.00198)x2 − 10.3(±1.15)x +397(±8.6), R2 = 0.914, RMSE = 8.7] and milk protein yield [�; y =−0.021(±0.0001)x2 + 20.9(±5.69)x + 681(±44.6), R2 = 0.833, RMSE =41.8] with MUN based on data derived from protein supplementationstudies (n = 188).
respect to environmental N emissions or reproductiveperformance).
Measurements of MUN appeared to provide an accu-rate prediction of urinary N excretion (Table 6), consis-tent with previous suggestions (Jonker et al., 1998;Kauffman and St-Pierre, 2001; Kohn et al., 2002). Pre-diction of the efficiency of N utilization and urinaryN excretion based on both MUN and milk yield wasassociated with RMSE of 19.3 g/kg N and 29 g/d, respec-tively. Use of mixed effects models improved the R2 andlowered the RMSE of these relationships, but both theintercept and coefficients of the models were relativelyunchanged when fixed models were used, suggestingthat the prediction may be sufficiently robust to be ap-plied in practice. Furthermore, these equations suggestthat changes in urinary N output in response to dietarychanges can be predicted very accurately (Table 6), par-ticularly when protein supplements are fed. Predictionbased on MUN alone indicated that urinary N excretionincreased 13.4 g/d per unit (mg/dL) increase in MUN.Corresponding equations reported by Jonker et al.(1998) and Kauffman and St-Pierre (2001) providesomewhat different slopes (12.2 and 17.2, respectively)and assume an intercept of zero. The model of Kauffmanand St-Pierre (2001) provides comparable estimates tothe prediction equation derived in this evaluation forMUN values of about 12 mg/dL, but further increasesin MUN could potentially overestimate urinary N excre-tion due to the assumption of a zero intercept (Figure3). Kohn et al. (2002) suggested that live weight shouldalso be included into the prediction of urinary N excre-
Journal of Dairy Science Vol. 87, No. 2, 2004
Figure 3. Comparison of the predictions of urinary N excretionbased on milk urea nitrogen (MUN) according to (◆) present data,(�) Jonker et al. (1998), (▲) Kauffman and St-Pierre (2001).
tion that may be in part analogous to the intercept ofthe relationship derived in this evaluation, wherebythe intercept corresponds to the constant effect of liveweight. However, live weight in the model of Kohn etal. (2002) appears to be proportional to MUN, not addi-tional as suggested in this study.
Alternatively, urinary N excretion can be predictedfrom MUN secretion. Overall, errors in the predictionwere numerically similar for MUNS compared withMUN. However, MUNS provided a more accurate pre-diction when diet composition was manipulated bychanges in the level or CP content of concentrate supple-ments. Predictions of urinary N excretion could be im-proved yet further if reliable measurements of dietaryCP content are available. This relationship clearly indi-cates that urinary N excretion is primarily governedby dietary CP content. However, the significant contri-bution of MUNS in the prediction of urinary N excretionis indicative of the importance of protein quality onboth the efficiency of N utilization for milk protein syn-thesis and magnitude of N losses in urine.
CONCLUSIONS
Dietary concentrations of CP and PBV were the mainnutritional factors influencing MUN. Associations withMUN were not improved by expressing dietary proteinas a function of energy. Nitrogen losses arising fromthe rumen or liberated during tissue AA catabolismhad similar effects on urea secretion in milk. Concentra-tions of MUN in milk provided a reliable estimate ofdietary CP content, whereas MUN or MUNS provided
NOUSIAINEN ET AL.396
an accurate prediction of urinary N excretion. On thebasis of this evaluation, measurement of MUN concen-tration or secretion can be used to monitor environmen-tal N emissions associated with milk production.
REFERENCES
Ahvenjarvi, S., A. Vanhatalo, and P. Huhtanen. 2002. Supple-menting barley or rapeseed meal to dairy cows fed grass-redclover silage: I. Rumen degradability and microbial flow. J.Anim. Sci. 80:2176–2187.
Ahvenjarvi, S., A. Vanhatalo, P. Huhtanen, and T. Varvikko. 1999.Effects of supplementation of a grass silage and barley diet withurea, rapeseed meal, and heat-moisture treated rapeseed cakeon milk production and omasal digesta flow in lactating dairycows. Acta Agric. Scand. Sect. A Anim. Sci. 49:179–189.
Annison, E. F. 1983. Metabolite utilization by the ruminant mam-mary gland. Pages 399–439 in Biochemistry of Lactation, Elsev-ier Science Publishers B.V., Amsterdam, The Netherlands.
Broderick, G. A., and M. K. Clayton. 1997. A statistical evaluationof animal and nutritional factors influencing concentrations ofmilk urea nitrogen. J. Dairy Sci. 80:2964–2971.
Carlsson, J., and B. Pehrson. 1994. The influence of the dietarybalance between energy and protein on milk urea concentration.Experimental trials assessed by two different protein evaluationsystems. Acta Vet. Scand. 35:193–205.
Castillo, A. R., E. Kebreab, D. E. Beever, and J. A. France. 2000. Areview of efficiency of nitrogen utilisation in lactating dairy cowsand its relationship with environmental pollution. J. Anim. FeedSci. 9:1–32.
De Peters, E. J., and J. D. Ferguson. 1992. Nonprotein nitrogen andprotein distribution in the milk of cows. J. Dairy Sci.75:3192–3209.
Frank, B., and C. Swensson. 2002. Relationship between content ofcrude protein in rations for dairy cows and milk yield, concentra-tion of urea in milk and ammonia emissions. J. Dairy Sci.85:1829–1838.
Godden, S. M., K. D. Lissemore, D. F. Kelton, J. H. Lumsden, K.E. Leslie, and J. S. Walton. 2000. Analytic validation of an infra-red milk urea assay and effects of sample acquisition factors onmilk urea results. J. Dairy Sci. 83:435–442.
Gustafsson, A. H., and J. Carlsson. 1993. Effects of silage quality,protein evaluation systems, and milk urea content on milk yieldand reproduction in dairy cows. Livest. Prod. Sci. 37:91–105.
Hof, G., M. D. Vervoorn, P. J. Lenaers, and S. Tamminga. 1997.Milk urea nitrogen as a tool to monitor protein nutrition of dairycows. J. Dairy. Sci. 80:3333–3340.
Huhtanen, P. 1998. Supply of nutrients and productive responsesin dairy cows given diets based on restrictively fermented silage.Agric. Food Sci. Finl. 7:219–250.
Huhtanen, P., H. Miettinen, and V. Toivonen. 1997. Effects of silagefermentation and postruminal casein supplementation in lactat-ing dairy cows. 1. Diet digestion and milk production. J. Sci.Food Agric. 74:450–458.
Jonker J. S., R. A. Kohn, and R. A. Erdman. 1998. Using milk ureanitrogen to predict nitrogen excretion and utilization efficiencyin lactating dairy cows. J. Dairy Sci. 81:2681–2692.
Jonker, J. S., R. A. Kohn, and J. High. 2002. Use of milk ureanitrogen to improve dairy cow diets. J. Dairy Sci. 85:939–946.
Kauffman, A. J., and N. R. St-Pierre. 2001. The relationship of milkurea nitrogen to urine nitrogen excretion in Holstein and Jerseycows. J. Dairy Sci. 84:2284–2294.
Khalili, H., and P. Huhtanen. 2002. Effect of casein infusion in therumen, duodenum, or both sites on factors affecting forage intakeand performance of dairy cows fed red clover-grass silage. J.Dairy Sci. 85:909–918.
Kirchgessner, M., M. Kreuzer, and D. A. Roth-Maier. 1986. Milkurea and protein content to diagnose energy and protein malnu-trition of dairy cows. Arch. Anim. Nutr. 36:192–197.
Journal of Dairy Science Vol. 87, No. 2, 2004
Kohn, R. A., K. F. Kalscheur, and E. Russek-Cohen. 2002. Evalua-tion of models to estimate urinary nitrogen and expected milkurea nitrogen. J. Dairy Sci. 85:227–233.
Littell, R. C., G. A. Milliken, W. W. Stroup, R. D. Wolfinger. 1996.SAS System for Mixed Models. SAS Inst. Inc., Cary, NC.
McCullough, H. 1967. The determination of ammonia in whole bloodby a direct calorimetric method. Clin. Chim. Acta 17:297–304.
Nousiainen, J., M. Rinne, M. Hellamaki, and P. Huhtanen. 2003.Prediction of the digestibility of the primary growth of grasssilages harvested at different stages of maturity from chemicalcomposition and pepsin cellulase solubility. Anim. Feed Sci.Technol. 103:97–111.
Oltner, R., and H. Wiktorsson. 1983. Urea concentrations in milkand blood as influenced by feeding various amounts of proteinand energy to dairy cows. Livest. Prod. Sci. 10:457–467.
Rajamaki, S., and A. Rauramaa. 1984. The automated determina-tion of urea on milk. Finn. Chem. Lett. 2:47.
Ropstad, E. L., L. Vik-Mo, and A. O. Refsdal. 1989. Levels of milkurea, plasma constituents, and rumen liquid ammonia in rela-tion to the feeding of dairy cows during early lactation. ActaVet. Scand. 30:199–208.
Roseler, D. K., J. D. Ferguson, C. J. Sniffen, and J. Herrema. 1993.Dietary protein degradability effects on plasma and milk ureanitrogen and milk nonprotein nitrogen in Holstein cows. J.Dairy. Sci. 76:525–534.
Schepers, A. J., and R. G. M. Meijer. 1998. Evaluation of the utiliza-tion of dietary nitrogen by dairy cows based on urea concetrationin milk. J. Dairy Sci. 81:579–584.
Shingfield, K. J., S. Jaakkola, and P. Huhtanen. 2001. Effects oflevel of nitrogen fertilizer application and various nitrogenoussupplements on milk production and nitrogen utilization of dairycows given grass silage-based diets. Anim. Sci. 73:541–554.
Shingfield, K. J., M. Jokela, K. Kaustell, P. Huhtanen, and J. Nousi-ainen. 1999. Association between protein feeding and reproduc-tive efficiency in the dairy cow: specific emphasis on proteinfeeding in Finland. Agric. Food Sci. Finl. 8:365–392.
Shingfield, K. J., A. Vanhatalo, and P. Huhtanen. 2003. Comparisonof heat-treated rapeseed expeller and solvent-extracted soya-bean meal as protein supplements for dairy cows fed grass silage-based diets. Anim. Sci. 77:305–317.
St-Pierre, N. R. 2001. Integrating quantitative findings from multi-ple studies using mixed model methodology. J. Dairy Sci.84:741–755.
Tilley, J. M. A., and R. A. Terry. 1963. A two-stage technique forin vitro digestion of forage crops. J. Br. Grassl. Soc. 18:104–111.
Tuori, M., K. Kuoppala, J. Valaja, E. Aimonen, E. Saarisalo, andP. Huhtanen. 2002. Rehutaulukot ja ruokintasuositukset 2002[web document]. Published 28.6.2002, [refereed 28.3.2003]. Webaccess at: http://www.agronet.fi/rehutaulukot/.
Vanhatalo, A., T. Varvikko, and P. Huhtanen. 2003. Effects of caseinand glucose on responses of cows fed diets based on restrictivelyfermented silage. J. Dairy Sci. 86:3260–3270.
Van Keulen, J., and B. A. Young. 1977. Acid insoluble ash as anatural marker for digestibility studies. J. Anim. Sci. 44:282–287.
APPENDIX 1
This list of publications describes data used to evalu-ate the effect of dietary and animal factors on milkurea nitrogen. Numbers in parenthesis indicate thenumber of mean treatment values derived from eachcitation.
Bertilsson J., R. J. Dewhurst, and M. Tuori. 2002. Effects of legumesilages on feed intake, milk production, and nitrogen efficiency.Landbauforschung Voelkenrode, SH 234:39–45. (20).
Heikkila, T. 1997. Kevat- ja syyssadon seka valkuais- ja vakirehuta-son vaikutus maidontuotannossa [The effect of grass silage har-
MILK UREA AS A PROTEIN DIAGNOSTIC 397
vest (primary vs. regrowth) and concentrate supplementationlevel on milk production]. Pages 183–186 in Proc. of Kotielaintie-teen paivat 1997, Maaseutukeskusten Liiton julkaisuja 914. (8).
Heikkila, T., and P. Huhtanen. 1995. Effects of physical treatmentof barley and rapeseed meal in dairy cows given grass silage-based diets. Agric. Food Sci. Finl. 5:399–412. (6).
Heikkila, T., S. Jaakkola, and P. Huhtanen. 2000. Effect of sodiumfertilization of cocksfoot leys on milk production. Pages 157–159in Grassland Science in Europe 5, Proc. of the 18th GeneralMeeting of the European Grassland Federation Aalborg, Den-mark 22–25 May 2000. (15).
Heikkila, T., S. Jaakkola, R. Sorunen-Cristian, and T. Mela. 2000.Yellow-flowered lucerne-grass silage in milk production. Pages353–355 in Grassland Science in Europe 5, Proc. of the 18thGeneral Meeting of the European Grassland Federation, Aal-borg, Denmark, 22–25 May 2000. (2).
Heikkila, T., and V. Toivonen. 1997. Herne ja rypsirouhe lehmienvalkuaisrehuna sailorehuruokinnalla [Pea and rapeseed mealas protein supplements for dairy cows fed grass silage baseddiets]. Pages 187–190 in Proc. of Kotielaintieteen paivat 1997,Maaseutukeskusten Liiton julkaisuja 914. (8).
Heikkila, T., V. Toivonen, and P. Huhtanen. 1998. Effects of andinteractions between the extent of silage fermentation and pro-tein supplementation in lactating dairy cows. Agric. Food Sci.Finl. 3:329–343. (6).
Heikkila, T., V. Toivonen, T. Keskinen, P. Parikka, and T. Tupasela.1999. Effects of two additives and harvesting methods of wiltedsilage and concentrate level on milk production. Pages 149–150in Proc. of the XIIth International Silage Conference, July 5–7,1999, Uppsala, Sweden. (8).
Heikkila, T., V. Toivonen, and T. Tupasela. 1997. Effect of additiveson big bale silage quality and milk production. Page 119 in J.A. M. van Arendonk, Editor-in-Chief, Book of Abstracts of the48th Annual Meeting of the European Association for AnimalProduction: Vienna, Austria, 25–28 August 1997. Wageningen,Wageningen Press. (3).
Huhtanen, P. 1993. The effects of concentrate energy source andprotein content on milk production in cows given grass silagead libitum. Grass and Forage Sci. 48:347–355. (4).
Huhtanen, P., S. Jaakkola, and P. Parssinen. 2001. Ruokonata lyp-sylehmien rehuna [Tall fescue as a feed for dairy cows]. Pages34–38 in (Editors) Niemelainen, O., M. Topi-Hulmi, and E. Saari-salo, Nurmitutkimusten satoa: tuloksia lannoituksesta, palko-kasveista, luomunurmista, laitumista, ruokonadasta. SuomenNurmiyhdistyksen julkaisu 14. (4).
Huhtanen, P., S. Jaakkola, and E. Saarisalo. 1995. The effects ofconcentrate energy source on the milk production of dairy cowsgiven a grass silage-based diet. Anim. Sci. 60:31–40. (8).
Huhtanen, P., S. Jaakkola, and J. Kylmanen. 1992. Tarkkelysrankinsuojauskasittelyn ja rypsilisan vaikutus lehmien tuotantoon[The effect of wet distillers grain treatment and protein supple-ment on milk production]. University of Helsinki, Departmentof Animal Science, Unpublished. (6).
Huhtanen, P., and H. Miettinen. 1992. Milk production and concen-trations of blood metabolites as influenced by the level of wetdistiller’s solubles in dairy cows receiving grass silage-baseddiet. Agric. Sci. Finl.1:279–290. (4).
Jaakkola, S., T. Heikkila, E. Saarisalo, and P. Huhtanen. 2002.Kokoviljasailorehun soveltuvuus lehmien ruokintaan [Wholecrop barley silage for milk production]. Pages 31–43 in Proc.(Editors) E. Saarisalo, and M. Topi-Hulmi, Rehuvaihtoehtojanautakarjatiloille: seminaari Jokioisilla 29.4.2002, SuomenNurmiyhdistyksen julkaisu 18. (20).
Jaakkola, S., and P. Huhtanen. 1998. Ohrarehun ja rankin kayttolypsylehmien ruokinnassa [The use of barley fiber and wet dis-tillers grain for dairy cows]. Agrifood Research Finland, AnimalNutrition. Unpublished. (8).
Jaakkola, S., and E. Joki-Tokola. 1999. Kokoviljasailorehun soveltu-vuus lypsylehman ruokintaan [Suitability of whole crop silagefor dairy cows]. Page E54 in Mita Suomi syo—ja milla hinnalla?Proc. of Agro-Food ′99, Tampere 2.-4.2.1999, Tampere-talo, Hel-sinki, Agro-Food ry/Agronomiliitto ry. (7)
Journal of Dairy Science Vol. 87, No. 2, 2004
Jaakkola, S., M. Rinne, T. Heikkila, V. Toivonen, and P. Huhtanen.1996. Muurahaishapon annostelutason, rypsirouheen ja propy-leeniglykolin vaikutukset maidontuotannossa [The effects ofrate of formic acid application for grass silage and supplementa-tion of the rapeseed meal and propylene glycol on milk produc-tion]. Pages 213–217 in Proc. Kotielaintieteen paivat 1996, Ko-tielaintiede 90 vuotta -juhlaseminaari. Maaseutukeskusten Lii-ton julkaisuja 905. (16)
Jaakkola, S., M. Rinne, K. Huuskonen, V. Vesterinen, and P. Huhta-nen. 1995. The effect of silage fermentation on the response toprotein supplementation in milk production and rumen fermen-tation in vitro. Page 68 in Proc. of the VII Symposium on ProteinMetabolism and Nutrition, 24–27 May 1995, Estacao ZootecnicaNacional, Portugal. (6)
Khalili, H., P. Huhtanen, K. Rinne, and M. Suvitie. 1997. The effectsof two levels of concentrates supplying the same amount of pro-tein on silage intake and milk production in cows given two grasssilages. Pages 3–4 in Proc. of the XVIII International GrasslandCongress, 8–19 June 1997, Winnipeg, Manitoba, Saskatoon, Sas-katchewan, Canada, Volume 2, Session 17 Forage Quality, IDNO. 194. (10).
Khalili, H., E. Kuusela, E. Saarisalo, and M. Suvitie. 1999. Use ofrapeseed and pea grain protein supplements for organic milkproduction. Agric. Food Sci. Finl. 8:239–252. (7)
Khalili, H., A. Sairanen, K. Hissa, and P. Huhtanen. 2001. Effectsof type and treatment of grain and protein source on dairy cowperformance. Anim. Sci. 72:573–584. (8)
Kokkonen, T., M. Tuori, V. Leivonen, and L. Syrjala-Qvist. 2000.Effect of silage dry matter content and rapeseed meal supple-mentation on dairy cows. 1. Milk production and feed utilisation.Anim. Feed Sci. Technol. 84:213–228. (4)
Nousiainen, J. I. 2001. [The effects of concentrate energy and proteinallocation on the milk production in dairy cows]. In Nousiainen,J. I. Maitotilojen tuotannonohjauksen ja seurannan kehittamis-projekti, Loppuraportti 1.1.1999–31.3.2001, 35 pages. AgrifoodResearch Finland, North-Savo Research Station. (4).
Rinne, M., S. Jaakkola, M. Jarvi, and P. Huhtanen. 1997. Effectsof gradual replacement of rapeseed cake with linseed cake in agrass silage-based diet for dairy cows. Agric. Food Sci. Finl.6:161–172. (4)
Rinne, M., S. Jaakkola, K. Kaustell T. Heikkila, and P. Huhtanen.1999. Silages harvested at different stages of grass growth v.concentrate foods as energy and protein sources in milk produc-tion. Anim. Sci. 69:251–263. (16)
Rinne, M., S. Jaakkola, T. Varvikko, and P. Huhtanen. 1999. Effectsof type and amount of rapeseed feed in milk production. ActaAgric. Scand. 49:137–148. (8)
Saarisalo, E., S. Jaakkola, and P. Huhtanen. 2002. Effects of supple-menting grass silage with protein on production of primiparouscows in late lactation. Pages 594–595 in Multi-function grass-lands: quality forages, animal products, and landscapes, Grass-land Science in Europe 7, Proc. of the 19th general meeting ofthe European Grassland Federation, La Rochelle, France, 27–30 May 2002. (8)
Saarisalo, E., S. Jaakkola, E. Skytta, and P. Huhtanen. 2002. Maito-happobakteeriymppien vaikutus esikuivattujen sailorehujenlaatuun ja lypsylehmien maidontuotantoon [The effects of inocu-lants on the silage quality and milk production]. Pages 71–74in Proc. of Maataloustieteen Paivat 2002, Kotielaintiede, 9.–10.1.2002 Viikki, Helsinki, Maaseutukeskusten Liiton julkai-suja 977. (8).
Saarisalo, E., S. Jaakkola, and R. Sormunen-Cristian. 1997 Sirppi-mailassailorehu lypsylehmien ruokinnassa [Yellow-flowered lu-cerne-grass silage in the feeding of dairy cows]. Pages 175–177in Proc. of Kotielaintieteen paivat 1997, MaaseutukeskustenLiiton julkaisuja 914. (2)
Sairanen, A. 2000. Lisavalkuaistaso lypsylehmien kannattavassaruokinnassa [Protein supplementation level and the profitabiltyof milk production]. Koetoiminta ja kaytanto 57(1):7 (8).
Sairanen, A., J. I. Nousiainen, and H. Khalili. 1999. Korkean vaki-rehumaaran vaikutus maitotuotokseen ja tuotannon kannatta-vuuteen [The effect of high concentrate level on the yield and
NOUSIAINEN ET AL.398
profitability of milk production]. Page P7 in Mita Suomi syo—ja milla hinnalla? Proc. of Agro-Food ′99, Tampere 2.–4.2.1999,Tampere-talo, Helsinki, Agro-Food ry/Agronomiliitto ry. (10).
Shingfield, K. J., S. Jaakkola, and P. Huhtanen. 2001. Effects oflevel of nitrogen fertilizer application and various nitrogenoussupplements on milk production and nitrogen utilization of dairycows given grass silage-based diets. Anim. Sci. 73:541–554. (8)
Shingfield, K. J., S. Jaakkola, and P. Huhtanen. 2002. Effect offorage conservation method, concentrate level and propylene
Journal of Dairy Science Vol. 87, No. 2, 2004
glycol on intake, feeding behaviour, and milk production of dairycows. Anim. Sci. 74:383–397. (16).
Shingfield, K. J., A. Vanhatalo, and P. Huhtanen. 2003. Comparisonof heat-treated rapeseed expeller and solvent-extracted soya-bean meal as protein supplements for dairy cows fed grass silage-based diets. Anim. Sci. 77:305–317.
Tuori, M. 1992. Rapeseed meal as a supplementary protein for dairycows, with the emphasis on the NORDIC AAT-PBV feed proteinevaluation system. Agric. Sci. Finl. 1:367–439. (18).
